Title: Rank-based Estimation for Linear and Mixed-effects Models

Abstract:

Rank-based estimation offers a robust, nonparametric alternative to traditional least squares or maximum-likelihood methods.  In addition to estimates of model parameters a complete inference --- including estimates of standard errors, tests of hypothesis, confidence intervals, and diagnostic procedures --- has been developed.  We have also developed R software packages for computation.  In this talk, we will provided a brief overview of rank-based procedures with applications to real data.

Bio:

John Kloke is a Visiting Assistant Professor at UCSB.  Prior to UCSB, Dr. Kloke was a faculty member at several schools --- the University of Wisconsin being the most recent.  He received his PhD in Statistics from Western Michigan University.  His research interests include nonparametrics and R package development.  Medical research is another area of focus --- especially the design and analysis of clinical trials and observational studies.